I've talked with many friends in a similar situation as yours (econ phd) and they all faced the same issues. Some advisors don't even care if you are coding in Python or in stone tables, as long as you "have three stars in your results table". Your comments about the quality of the computations also reminded me a lot about Ken Judd's (often ignored) arguments about paying attention to all these implementation details.
Good luck with your current goals, and I really hope you are more successful than within the limitations and issues of academia.
PS: Just out of curiosity, what were you interested on, regarding your IO research?
>>> Some advisors don't even care if you are coding in Python or in stone tables, as long as you "have three stars in your results table".
That is beautiful. It perfectly describes the situation!
I still to this day have scripts that collect data on a variety of markets, but I tend to favor applications to platforms and multi-sided markets. My main one was mobile apps and app stores, but I also am researching video games, hotels, desktop CPUs, to name a few. The questions that interest me most are antitrust problems and finding creative ways to measure competition in an Internet market. I also enjoy IO theory, with my absolute favorite topic being low-price guarantees. To this day, I'm still digging for a creative way to do an empirical study on low-price guarantees to test a theoretical hypothesis I wrote a paper about in an IO theory course I took in the 1st semester of my PhD (was the only 1st year student to take a field course in year 1).
In the first semester of my 2nd year of my PhD (left after year 2), I got my solo work into a conference held jointly by Harvard and MIT called the "First Cambridge Area Economics and Computation Day", but my advisors didn't seemed to care much. The conference was awesome and my work got a lot of attention there, including a long talk with the Chief Economist at Microsoft.
Trying to understand how firms compete in the android store actually sounds like a great topic. Of course, we can only observe--at most--prices and sales per product, but it would be nice to think of how firms compete between each other and their reactions. My guess is that the biggest problem would be to find an identification strategy, like a change in the structure of the app store, so we can exogenize our regressors.
BTW, great job with AppNash, it looks very interesting!
If you want to talk more about the econometric approach, feel free to email me. Would love to chat more. If you discovered AppNash, then you can also figure out my email :). FWIW, your point about finding exogenous changes is right on (model = 2SLS), however getting the quantity sold data is not a simple procedure -- it's a matter of converting sales ranks to quantities. With out giving away all the secrets in public, here's a classic paper that guided some of my approach: http://www.business.illinois.edu/finance/papers/2003/chevali...
The original intent for my dissertation was the following: Chapter 1: Describe the theoretical framework of how to set up a system to automate the collection of data from an entire market (enabling the safe use of population estimators); Chapter 2: Apply that theory to analyze the app store data to test various firm- and product-level questions about the structure of strategies taken by firms (developers) in the app stores; Chapter 3: Another application of the theory using higher frequency data from the Internet (probably hotels, since hotels change their rates at the minutely/hourly level quite often, and regional hotel markets provide very well defined markets so new entry takes a while and is thus easy to account for).
Good luck with your current goals, and I really hope you are more successful than within the limitations and issues of academia.
PS: Just out of curiosity, what were you interested on, regarding your IO research?